// Test for case with import and find ops.
func.func @hashtable_import_then_find(%arg0: tensor<5x!tf_type.string>) -> tensor<*xi64> {
  %cst = arith.constant dense<["emerson", "lake", "palmer"]> : tensor<3x!tf_type.string>
  %cst_0 = arith.constant dense<-1> : tensor<i64>
  %cst_1 = arith.constant dense<[0, 1, 2]> : tensor<3xi64>

func.func private @interleaved_tf_and_mhlo() {
  %m0 = mhlo.constant dense<0> : tensor<i32>
  %t0 = "tf.Const"() { value = dense<0> : tensor<1xi32> } : () -> tensor<1xi32>
  %m1 = mhlo.constant dense<1> : tensor<i32>
  %t1 = "tf.Const"() { value = dense<1> : tensor<1xi32> } : () -> tensor<1xi32>
  %m2 = mhlo.constant dense<1> : tensor<i32>
  %t2 = "tf.Const"() { value = dense<1> : tensor<1xi32> } : () -> tensor<1xi32>

    StringRef calibration_data_dir,
    const std::vector<std::string>& aggregator_ops_to_ignore);

// Adds generated pass default constructors or options definitions.
#define GEN_PASS_DECL
// Adds generated pass registration functions.
#define GEN_PASS_REGISTRATION
#include "tensorflow/compiler/mlir/quantization/stablehlo/passes/passes.h.inc"

}  // namespace mlir::quant::stablehlo

  // CHECK-DAG: %[[OFFSET_BCAST:.+]] = "mhlo.broadcast_in_dim"(%[[OFFSET]]) <{broadcast_dimensions = dense<1> : tensor<1xi64>}> : (tensor<256xf32>) -> tensor<4x256xf32>
  // CHECK-DAG: %[[MEAN_BCAST:.+]] = "mhlo.broadcast_in_dim"(%[[MEAN]]) <{broadcast_dimensions = dense<1> : tensor<1xi64>}> : (tensor<256xf32>) -> tensor<4x256xf32>

It is provided directly by Starlette, but you can import it from `fastapi`:

```python
from fastapi import WebSocket
```

!!! tip
    When you want to define dependencies that should be compatible with both HTTP and WebSockets, you can define a parameter that takes an `HTTPConnection` instead of a `Request` or a `WebSocket`.

::: fastapi.WebSocket
    options:
        members:
            - scope
            - app

// CHECK-LABEL:   func.func @prune_unused_restore
func.func @prune_unused_restore() {
  %cst = "tf.Const"() <{value = dense<"restore_ariables"> : tensor<!tf_type.string>}> : () -> tensor<!tf_type.string>
  %cst_0 = "tf.Const"() <{value = dense<""> : tensor<1x!tf_type.string>}> : () -> tensor<1x!tf_type.string>
  %cst_1 = "tf.Const"() <{value = dense<"y"> : tensor<1x!tf_type.string>}> : () -> tensor<1x!tf_type.string>
  // CHECK-NOT: tf.RestoreV2

  // CHECK: func @f()
  func.func @f(%arg0: tensor<!tf_type.resource<tensor<f32>>> {tf_saved_model.bound_input = @v})
  attributes {tf_saved_model.exported_names = ["f"]} {
    %val = "tf.ReadVariableOp"(%arg0) : (tensor<!tf_type.resource<tensor<f32>>>) -> tensor<f32>
    // CHECK: "tf.Const"() <{value = dense<1.000000e+00> : tensor<f32>}>
    func.return
  }

// due to the calling conventions and initialization of constants.

//go:build gc && !purego

#include "textflag.h"

#define OUT  R3
#define INP  R4
#define LEN  R5
#define KEY  R6
#define CNT  R7
#define TMP  R15

#define CONSTBASE  R16
#define BLOCKS R17

// for VPERMXOR
#define MASK  R18

DATA consts<>+0x00(SB)/8, $0x3320646e61707865
DATA consts<>+0x08(SB)/8, $0x6b20657479622d32

// H6 = g + H6
// H7 = h + H7

// Wt = Mt; for 0 <= t <= 15
#define MSGSCHEDULE0(index) \
	MOVL	(index*4)(SI), AX; \
	BSWAPL	AX; \
	MOVL	AX, (index*4)(BP)

// Wt = SIGMA1(Wt-2) + Wt-7 + SIGMA0(Wt-15) + Wt-16; for 16 <= t <= 63
//   SIGMA0(x) = ROTR(7,x) XOR ROTR(18,x) XOR SHR(3,x)
//   SIGMA1(x) = ROTR(17,x) XOR ROTR(19,x) XOR SHR(10,x)
#define MSGSCHEDULE1(index) \
	MOVL	((index-2)*4)(BP), AX; \
	MOVL	AX, CX; \

    attributes {tf_saved_model.exported_names = ["lowers_to_stablehlo"]}
  {
    // CHECK-DAG: [[one:%.*]] = stablehlo.constant dense<1>
    // CHECK-DAG: [[twenty:%.*]] = stablehlo.constant dense<20>
    // CHECK-DAG: [[r3:%.*]] = stablehlo.subtract [[twenty]], %arg0
    // CHECK-DAG: [[zero:%.*]] = stablehlo.constant dense<0>
    // CHECK-DAG: [[r4:%.*]] = stablehlo.divide [[r3]], [[one]]
    // CHECK-DAG: [[r5:%.*]] = stablehlo.compare NE